Micron CEO Sold Before the Best Quarter Ever|9x Forward PE Says Market Still Doesnt Believe It

The Quarter Bears Said Was Already Priced In

Micron Technology reported fiscal Q3 revenue of $41.46 billion on June 24, 2026 — a 345% increase from one year earlier, beating analyst estimates of $36 billion by 15%. Adjusted earnings of $25.11 per share topped the $20.78 consensus. The company then guided Q4 revenue of $50 billion, against a Street estimate of $43 billion. Those numbers did not come from a company coasting on a favorable cycle. They came the day after Micron had fallen 13% on fears that SK Hynix was slowing its expansion — a signal the market read as evidence that AI memory demand was cooling. The bears had four specific arguments going into earnings: gross margins were peaking, capital expenditure was masking profitability, customer concentration created fragility, and the valuation already priced in perfection. Every one of those arguments was disproven in the same session. GAAP gross margin expanded to 84.6% from 37.7% a year earlier, with Q4 guided to approximately 86%. Free cash flow reached $18.3 billion — roughly equal to what Micron generated in all of fiscal 2025 as revenue. Long-term supply agreements now number 16, with hyperscalers, automakers, and AI infrastructure companies locked in for three to five years. The stock rose 16% the next day. That sequence — a 13% pre-earnings selloff followed by a 16% post-earnings surge — is the surface story. But the surface story misses the more difficult question the numbers raised. If every bear argument was wrong, what is the one signal that still points in the opposite direction from the results?

The CEO Sold 37,435 Shares Before the Best Quarter in Company History

Before reporting a quarter that beat by 15% and guided the next quarter 16% above consensus, Micron CEO Sanjay Mehrotra sold 37,435 shares at approximately $979.37. The sale was pre-scheduled, which limits its informational value in isolation. But the interpretation problem runs deeper than that. Mehrotra did not merely preside over strong results — he simultaneously announced 16 long-term Strategic Customer Agreements that lock buyers in for three to five years. Those contracts are not a hedge against uncertainty. They are a deliberate structural choice to sacrifice peak-cycle upside in exchange for durability. A CEO who believed memory was approaching a cyclical peak would not be signing binding five-year supply commitments with hyperscalers and automakers. He would be taking spot revenue at the highest prices the market will bear. The act of locking customers into multi-year contracts — with concrete volume commitments and pricing floors — is the opposite of a company preparing for a downturn. That is the participant conflict the pool surfaces directly. One class — the CEO, selling pre-results shares — is behaving as if the stock has captured most of the upside. A different class — the hyperscalers signing 3-to-5-year supply agreements — is behaving as if the constraint persists far beyond the current price level. Those are not two interpretations of the same fact. They are two actors taking opposite financial positions on the same underlying reality: the durability of AI memory demand. The question the investor must answer is not whether the quarter was good — it was. The question is which actor's revealed position better forecasts the next 18 months: the executive trimming near the peak, or the hyperscaler locking in supply through a decade of contract obligations.

Why the Shortage Cannot Be Fixed Before 2028

The supply side of this conflict is where CEO Mehrotra's forward language becomes most significant. On the earnings call, he stated that AI-driven demand for DRAM and NAND "significantly" exceeds supply, and will continue to do so "beyond calendar 2027." He added that even gradual improvement in industry supply expected in 2028 leaves no "line of sight" on when memory supply will catch up with demand. That is not a promotional statement — it is structural. HBM, the high-bandwidth memory used in AI accelerators, consumes roughly three times the wafer capacity of conventional DRAM per bit. Every wafer Micron shifts to HBM removes supply from standard memory markets. New fab capacity requires years to build and billions of dollars in greenfield construction — brownfield modifications cannot close the gap fast enough. Micron has already shipped HBM4 in high volume for its lead customer's platform and has sent HBM4E qualification samples built on its next-generation 1-gamma DRAM node. But even with those shipments running, the constraint is structural, not operational: skilled worker shortages, long fab construction lead times, complex regulatory dynamics, and the need for enhanced energy infrastructure form a set of bottlenecks that no single management decision can remove within two years. The hidden assumption the bear case requires — that supply can respond fast enough to normalize margins before the current stock price becomes irrational — is not supported by the physical constraints of fab construction timelines. The 16 multi-year customer agreements are the structural evidence that Micron's largest buyers have already conducted that analysis and concluded the shortage persists long enough to justify legally binding commitments. The bears are correct that memory cycles are brutal. They are wrong if they assume this cycle resolves on normal timelines.

2x Forward P/E — Why the Market Prices MU Like a Utility, and What Changes That

Micron shares currently trade at a forward price-to-earnings multiple of 9.2x — roughly the same level as Verizon and AT&T, and well below the S&P 500's 20.4x weighted average. That discount is not irrational. It reflects the memory industry's history: Micron posted an annual loss as recently as fiscal 2023, when a supply glut collapsed margins and revenue fell 49% year over year. The 9.2x multiple contains a probability-weighted view of a future mean-reversion — the market is discounting the possibility that what happened in 2023 will happen again. That is the buried assumption driving the valuation gap. The structural counter-evidence arrived in the same earnings report. Apple, according to the CNBC market recap, is now raising prices on iPhones, iPads, and MacBooks because of a shortage of memory chips and storage. The same constraint driving Micron's 84.6% gross margin is flowing through into consumer device retail prices — a transmission path that does not occur in a commodity glut. A one-off earnings beat does not reprice that multiple. What could reprice it is a sequence of quarters that demonstrates the long-term supply agreements are delivering the durable, predictable revenue they were designed to provide. The counter-evidence worth monitoring: Paul Meeks of Freedom Capital Markets noted before the report that revenue growth is projected to slow to 76% in fiscal 2027 and 7% in 2028. If that deceleration arrives earlier or more sharply than the contract structure implies, the 9.2x multiple will look less like a discount and more like a fair value. The Q4 revenue print — guided at $50 billion — due in September 2026 is the earliest leading variable that decides the question. A delivery at $50 billion with gross margin at or above 85% is the signal that the multi-year contract structure is reshaping the cycle, and the 9.2x multiple is a genuine anomaly. A miss below $47 billion, or margin contraction below 82%, is the signal that the bear case — cyclical forces overriding contract structure — retains its explanatory power. The holder who entered on the 13% pre-earnings selloff should watch that September print as the first real test. The watcher who has not yet entered faces the same binary: the same $50 billion delivery is what converts the 9.2x discount from a rational risk premium into an entry signal.